I run a small site, buyanov.io, that helps people get into AI without drowning. Six months in, I have a private channel, a public one, a paid offer, and a few hundred real customers. I am not going to give you a growth story. I am going to give you what the numbers actually showed.
The 60% rule
Roughly 60% of people who contact me don't want to learn AI. They want one specific thing done — a difficult message, a legal letter, a meeting plan, a half-written text. They pay for the result, not the skill. The skill would have been free. They are not paying for free. They are paying for "no longer on my desk".
If you are building anything in the AI space, internalize this: most of the demand is not "teach me" and not "do it for me". It is "I have decided I will not be the one to do this part". The shape of that offering is consulting, not course.
What "fluency" actually means here
I use "operator" rather than "user" on purpose. A user consumes an app. An operator runs a process that includes AI. The difference matters because a user buys a subscription, and an operator buys a workflow.
In practice, the people I work with are not asking for a better prompt. They are asking for a better morning. They want their inbox triaged, their week planned, their draft rewritten, their kid's school task explained. The unit of value is the morning, not the chat.
What I stopped doing
I stopped teaching frameworks. Frameworks are easy to write and hard to apply. I teach one specific thing at a time, for one specific job, with one specific pass/fail criterion.
The pass/fail criterion is the load-bearing part. Without it, a session is "I learned something". With it, the session ends in one of two states: it worked, or it didn't, and now we know which. That second state is the one I want to be in. The first state is fine but it doesn't generate the next call.
What I keep doing
I keep the price low. I keep the offering small. I keep the door open to the same person coming back for a different one specific thing. The funnel is: one thing → trust → another thing → trust → eventually a method. Not the other way around. If you try to sell a method first, you are selling uncertainty.
The number that matters
The number that matters is not MRR. It is the percentage of sessions that end with "I would do this again for the same price". I do not track that percentage formally. I do not need to. I can feel it. It is high enough that I am still here.